This may cause an uproar:
A local court in Germany has made a landmark ruling that could significantly impact the operation of search engines and AI chatbots worldwide. The Munich Regional Court has determined that Google is liable for false statements generated by its AI Overviews feature. This ruling compels Google to take responsibility for preventing the circulation of inaccurate claims through its search engine.
The case originated when two publishers found that Google’s AI-generated summaries linked them to dubious business practices and scams, without any factual basis. These publishers sent a cease-and-desist letter to Google earlier in the year. However, Google refuted the accusations, arguing that its AI tool provides warnings that the information may be erroneous and should be independently verified.
The court’s analysis showed that Google’s AI synthesized information from various sources, combining it in a way that falsely associated the plaintiffs with illicit activities. Unlike traditional search engines that merely list links, Google’s AI tool created “independent, new, and substantial statements” based on misinterpreted data. [IT Magazine]
When a service collides with individual reputational concerns, when it spreads falsehoods, which is more important, this sometimes-wrong service, or the reputation?
The nature of truth and associated questions, such as epistemology, has challenged philosophers for millenia, and scientists wrestling with data collection, along with baffling cause & effect scenarios such as placebos and nocebos, for centuries. Distinguishing a measurement from someone jiggling the probe is sometimes a difficult business, for example.
Google is providing a generative AI service, which means they’ve processed the content of some part of the Web and use that to generate their service contents. It should go without saying that the Web isn’t trustworthy as a general rule, and even trusty parts are sometimes temporarily corrupted. Does Google’s quest for yet larger profits adequate justification for putting folks’ reputations at risk, for distributing fallacious information for possibly critical services?
Or is this an early incident in the fall of the generative AI services?
